In [1]:
import os
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

import seaborn as sns
sns.set_context('notebook', font_scale=1.5, rc={'lines.markeredgewidth': 2})
In [2]:
%load_ext autoreload
%autoreload 2

%matplotlib inline
In [3]:
from visual_behavior.data_access import loading as loading

from allensdk.brain_observatory.behavior.behavior_project_cache import VisualBehaviorOphysProjectCache
In [4]:
import visual_behavior.visualization.utils as utils

import visual_behavior.data_access.utilities as utilities

import visual_behavior.visualization.ophys.platform_paper_figures as ppf

load experiments_table from cache

In [5]:
cache_dir = loading.get_platform_analysis_cache_dir()
cache = VisualBehaviorOphysProjectCache.from_s3_cache(cache_dir=cache_dir)
# print(cache_dir)
In [6]:
experiments_table = loading.get_platform_paper_experiment_table()
print(len(experiments_table))

1249

add useful columns

In [7]:
experiments_table = utilities.add_cell_type_column(experiments_table)
experiments_table = utilities.add_experience_exposure_column(experiments_table)
experiments_table = utilities.add_binned_depth_column(experiments_table)

get useful info

In [8]:
experience_level = np.sort(experiments_table.experience_level.unique())
print(experience_level)

['Familiar' 'Novel 1' 'Novel >1']
In [9]:
exposures = np.sort(experiments_table.experience_exposure.unique())
print(exposures)

['Familiar 0' 'Familiar 1' 'Familiar 2' 'Familiar 3' 'Familiar > 3'
 'Novel 0' 'Novel 1' 'Novel 2' 'Novel 3' 'Novel > 3']
In [10]:
cell_types = np.sort(experiments_table.cell_type.unique() )
print(cell_types)

['Excitatory' 'Sst Inhibitory' 'Vip Inhibitory']

get multi_session_df

In [11]:
%%time

# set various params
df_name = 'trials_response_df'
conditions = ['cell_specimen_id', 'stimulus_change', 'epoch']
use_events = True
filter_events = False

# load multi_session_df
multi_session_df = loading.get_multi_session_df(cache_dir, df_name, conditions, experiments_table, 
                                                use_events=use_events, filter_events=filter_events)
print(len(multi_session_df.ophys_experiment_id.unique()))

original_multi_session_df = multi_session_df.copy()

# keep changes only
multi_session_df = multi_session_df[multi_session_df.stimulus_change==True]

# limit to platform paper dataset
multi_session_df = multi_session_df[multi_session_df.ophys_experiment_id.isin(experiments_table.index.values)]
print(len(multi_session_df.ophys_experiment_id.unique()))

# merge with metadata
multi_session_df = multi_session_df.merge(experiments_table, on='ophys_experiment_id')
print(len(multi_session_df.ophys_experiment_id.unique()))

multi_session_df = multi_session_df.reset_index()

1305
1249
1249
Wall time: 1min 11s

get rid of weird 7th epoch

In [12]:
multi_session_df = multi_session_df[multi_session_df.epoch!=6] # its the 6th index

add experience_epoch column

In [13]:
def merge_experience_epoch(row):
    return row.experience_level+' epoch '+str(int(row.epoch)+1)
In [14]:
multi_session_df['experience_epoch'] = multi_session_df[['experience_level', 'epoch']].apply(axis=1, func=merge_experience_epoch)

experience_epoch = np.sort(multi_session_df.experience_epoch.unique())
experience_epoch
Out[14]:
array(['Familiar epoch 1', 'Familiar epoch 2', 'Familiar epoch 3',
       'Familiar epoch 4', 'Familiar epoch 5', 'Familiar epoch 6',
       'Novel 1 epoch 1', 'Novel 1 epoch 2', 'Novel 1 epoch 3',
       'Novel 1 epoch 4', 'Novel 1 epoch 5', 'Novel 1 epoch 6',
       'Novel >1 epoch 1', 'Novel >1 epoch 2', 'Novel >1 epoch 3',
       'Novel >1 epoch 4', 'Novel >1 epoch 5', 'Novel >1 epoch 6'],
      dtype=object)

before any filtering

In [15]:
df = multi_session_df.copy()
df = df.reset_index()
In [16]:
utilities.count_mice_expts_containers_cells(df)
Out[16]:
n_mouse_id n_ophys_experiment_id n_ophys_container_id n_cell_specimen_id
cell_type experience_level
Excitatory Familiar 41 280 88 18782
Novel 1 36 71 71 9225
Novel >1 41 196 88 15705
Sst Inhibitory Familiar 19 144 46 738
Novel 1 15 37 37 414
Novel >1 19 107 45 678
Vip Inhibitory Familiar 22 217 69 1852
Novel 1 17 47 47 792
Novel >1 22 150 66 1547
In [17]:
fig, ax = plt.subplots(figsize=(8,5))
ax2 = ax.twinx()
cre_line_colors = utils.get_cre_line_colors()
ax2 = sns.pointplot(data=df[df.cell_type==cell_types[0]], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors, hue_order=cell_types, ax=ax2)
ax2.set_ylabel('excitatory mean response')
ax2.legend(bbox_to_anchor=(1.6,1))
ax2.set_ylim(ymin=0)
ax = sns.pointplot(data=df[df.cell_type.isin(cell_types[1:])], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors[1:], hue_order=cell_types[1:], ax=ax)
ax.set_ylabel('inhibitory mean response')
ax.set_ylim(ymin=0)
# ax.legend(bbox_to_anchor=(1.2,1))
ax.get_legend().remove()
ax.set_xticklabels(experience_epoch, rotation=90);
ax.set_title('average change evoked response over time')
ax.vlines(x=5.5, ymin=0, ymax=0.04, color='gray', linestyle='--')
ax.vlines(x=11.5, ymin=0, ymax=0.04, color='gray', linestyle='--')
Out[17]:
<matplotlib.collections.LineCollection at 0x2424cb8eb70>

limit to containers with all 3 experience levels

In [18]:
df = multi_session_df.copy()
df = utilities.limit_to_containers_with_all_experience_levels(df)
In [19]:
utilities.count_mice_expts_containers_cells(df)
Out[19]:
n_mouse_id n_ophys_experiment_id n_ophys_container_id n_cell_specimen_id
cell_type experience_level
Excitatory Familiar 36 227 71 15268
Novel 1 36 71 71 9225
Novel >1 36 142 71 11962
Sst Inhibitory Familiar 15 115 37 598
Novel 1 15 37 37 414
Novel >1 15 85 37 548
Vip Inhibitory Familiar 17 138 44 1233
Novel 1 17 44 44 742
Novel >1 17 96 44 1067
In [20]:
fig, ax = plt.subplots(figsize=(8,5))
ax2 = ax.twinx()
cre_line_colors = utils.get_cre_line_colors()
ax2 = sns.pointplot(data=df[df.cell_type==cell_types[0]], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors, hue_order=cell_types, ax=ax2)
ax2.set_ylabel('excitatory mean response')
ax2.legend(bbox_to_anchor=(1.6,1))
ax2.set_ylim(ymin=0)
ax = sns.pointplot(data=df[df.cell_type.isin(cell_types[1:])], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors[1:], hue_order=cell_types[1:], ax=ax)
ax.set_ylabel('inhibitory mean response')
ax.set_ylim(ymin=0)
# ax.legend(bbox_to_anchor=(1.2,1))
ax.get_legend().remove()
ax.set_xticklabels(experience_epoch, rotation=90);
ax.set_title('average change evoked response over time')
ax.vlines(x=5.5, ymin=0, ymax=0.04, color='gray', linestyle='--')
ax.vlines(x=11.5, ymin=0, ymax=0.04, color='gray', linestyle='--')
Out[20]:
<matplotlib.collections.LineCollection at 0x2423a73fbe0>

limit to containers with all 3 levels for closest active sessions

In [21]:
df = multi_session_df.copy()
df = utilities.limit_to_last_familiar_second_novel_active(df)
df = utilities.limit_to_containers_with_all_experience_levels(df)
In [22]:
utilities.count_mice_expts_containers_cells(df)
Out[22]:
n_mouse_id n_ophys_experiment_id n_ophys_container_id n_cell_specimen_id
cell_type experience_level
Excitatory Familiar 34 62 62 7302
Novel 1 34 62 62 7974
Novel >1 34 62 62 7544
Sst Inhibitory Familiar 15 30 30 321
Novel 1 15 30 30 337
Novel >1 15 30 30 320
Vip Inhibitory Familiar 17 42 42 794
Novel 1 17 42 42 718
Novel >1 17 42 42 765
In [23]:
fig, ax = plt.subplots(figsize=(8,5))
ax2 = ax.twinx()
cre_line_colors = utils.get_cre_line_colors()
ax2 = sns.pointplot(data=df[df.cell_type==cell_types[0]], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors, hue_order=cell_types, ax=ax2)
ax2.set_ylabel('excitatory mean response')
ax2.legend(bbox_to_anchor=(1.6,1))
ax2.set_ylim(ymin=0)
ax = sns.pointplot(data=df[df.cell_type.isin(cell_types[1:])], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors[1:], hue_order=cell_types[1:], ax=ax)
ax.set_ylabel('inhibitory mean response')
ax.set_ylim(ymin=0)
# ax.legend(bbox_to_anchor=(1.2,1))
ax.get_legend().remove()
ax.set_xticklabels(experience_epoch, rotation=90);
ax.set_title('average change evoked response over time')
ax.vlines(x=5.5, ymin=0, ymax=0.04, color='gray', linestyle='--')
ax.vlines(x=11.5, ymin=0, ymax=0.04, color='gray', linestyle='--')
Out[23]:
<matplotlib.collections.LineCollection at 0x2420dd28d68>

limit to matched cells in all 3

In [24]:
df = multi_session_df.copy()
df = utilities.limit_to_cell_specimen_ids_matched_in_all_experience_levels(df)
In [25]:
utilities.count_mice_expts_containers_cells(df)
Out[25]:
n_mouse_id n_ophys_experiment_id n_ophys_container_id n_cell_specimen_id
cell_type experience_level
Excitatory Familiar 36 226 71 5695
Novel 1 36 71 71 5695
Novel >1 36 142 71 5695
Sst Inhibitory Familiar 15 112 36 292
Novel 1 15 36 36 292
Novel >1 15 84 36 292
Vip Inhibitory Familiar 17 137 44 540
Novel 1 17 44 44 540
Novel >1 17 96 44 540
In [26]:
fig, ax = plt.subplots(figsize=(8,5))
ax2 = ax.twinx()
cre_line_colors = utils.get_cre_line_colors()
ax2 = sns.pointplot(data=df[df.cell_type==cell_types[0]], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors, hue_order=cell_types, ax=ax2)
ax2.set_ylabel('excitatory mean response')
ax2.legend(bbox_to_anchor=(1.6,1))
ax2.set_ylim(ymin=0)
ax = sns.pointplot(data=df[df.cell_type.isin(cell_types[1:])], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors[1:], hue_order=cell_types[1:], ax=ax)
ax.set_ylabel('inhibitory mean response')
ax.set_ylim(ymin=0)
# ax.legend(bbox_to_anchor=(1.2,1))
ax.get_legend().remove()
ax.set_xticklabels(experience_epoch, rotation=90);
ax.set_title('average change evoked response over time')
ax.vlines(x=5.5, ymin=0, ymax=0.04, color='gray', linestyle='--')
ax.vlines(x=11.5, ymin=0, ymax=0.04, color='gray', linestyle='--')
Out[26]:
<matplotlib.collections.LineCollection at 0x2420ff19208>

limit to matched cells in closest active sessions

In [27]:
df = multi_session_df.copy()
df = utilities.limit_to_last_familiar_second_novel_active(df)
df = utilities.limit_to_cell_specimen_ids_matched_in_all_experience_levels(df)
In [28]:
utilities.count_mice_expts_containers_cells(df)
Out[28]:
n_mouse_id n_ophys_experiment_id n_ophys_container_id n_cell_specimen_id
cell_type experience_level
Excitatory Familiar 34 62 62 3326
Novel 1 34 62 62 3326
Novel >1 34 62 62 3326
Sst Inhibitory Familiar 14 28 28 200
Novel 1 14 28 28 200
Novel >1 14 28 28 200
Vip Inhibitory Familiar 17 41 41 415
Novel 1 17 41 41 415
Novel >1 17 41 41 415
In [29]:
fig, ax = plt.subplots(figsize=(8,5))
ax2 = ax.twinx()
cre_line_colors = utils.get_cre_line_colors()
ax2 = sns.pointplot(data=df[df.cell_type==cell_types[0]], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors, hue_order=cell_types, ax=ax2)
ax2.set_ylabel('excitatory mean response')
ax2.legend(bbox_to_anchor=(1.6,1))
ax2.set_ylim(ymin=0)
ax = sns.pointplot(data=df[df.cell_type.isin(cell_types[1:])], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors[1:], hue_order=cell_types[1:], ax=ax)
ax.set_ylabel('inhibitory mean response')
ax.set_ylim(ymin=0)
# ax.legend(bbox_to_anchor=(1.2,1))
ax.get_legend().remove()
ax.set_xticklabels(experience_epoch, rotation=90);
ax.set_title('average change evoked response over time')
ax.vlines(x=5.5, ymin=0, ymax=0.04, color='gray', linestyle='--')
ax.vlines(x=11.5, ymin=0, ymax=0.04, color='gray', linestyle='--')
Out[29]:
<matplotlib.collections.LineCollection at 0x242238ae198>
In [ ]:
In [ ]:
In [ ]:
In [ ]:

for V1 or LM only

In [30]:
data = df.copy()
data = data[data.targeted_structure=='VISl']


fig, ax = plt.subplots(figsize=(8,5))
ax2 = ax.twinx()
cre_line_colors = utils.get_cre_line_colors()
ax2 = sns.pointplot(data=data[data.cell_type==cell_types[0]], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors, hue_order=cell_types, ax=ax2)
ax2.set_ylabel('excitatory mean response')
ax2.legend(bbox_to_anchor=(1.6,1))
ax2.set_ylim(ymin=0)
ax = sns.pointplot(data=data[data.cell_type.isin(cell_types[1:])], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors[1:], hue_order=cell_types[1:], ax=ax)
ax.set_ylabel('inhibitory mean response')
ax.set_ylim(ymin=0)
# ax.legend(bbox_to_anchor=(1.2,1))
ax.get_legend().remove()
ax.set_xticklabels(experience_epoch, rotation=90);
ax.set_title('average change evoked response over time')
ax.vlines(x=5.5, ymin=0, ymax=0.05, color='gray', linestyle='--')
ax.vlines(x=11.5, ymin=0, ymax=0.05, color='gray', linestyle='--')
Out[30]:
<matplotlib.collections.LineCollection at 0x24212dfa4e0>
In [ ]:

restrict to mesoscope

In [31]:
data = df.copy()
data = data[data.project_code=='VisualBehaviorMultiscope']


fig, ax = plt.subplots(figsize=(8,5))
ax2 = ax.twinx()
cre_line_colors = utils.get_cre_line_colors()
ax2 = sns.pointplot(data=data[data.cell_type==cell_types[0]], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors, hue_order=cell_types, ax=ax2)
ax2.set_ylabel('excitatory mean response')
ax2.legend(bbox_to_anchor=(1.6,1))
ax2.set_ylim(ymin=0)
ax = sns.pointplot(data=data[data.cell_type.isin(cell_types[1:])], x='experience_epoch', y='mean_response', hue='cell_type', 
                   order=experience_epoch, palette=cre_line_colors[1:], hue_order=cell_types[1:], ax=ax)
ax.set_ylabel('inhibitory mean response')
ax.set_ylim(ymin=0)
# ax.legend(bbox_to_anchor=(1.2,1))
ax.get_legend().remove()
ax.set_xticklabels(experience_epoch, rotation=90);
ax.set_title('average change evoked response over time')
ax.vlines(x=5.5, ymin=0, ymax=0.02, color='gray', linestyle='--')
ax.vlines(x=11.5, ymin=0, ymax=0.02, color='gray', linestyle='--')
Out[31]:
<matplotlib.collections.LineCollection at 0x24232cc0e80>
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:

heatmap of all mice epoch level responses

In [32]:
df = multi_session_df.copy()
In [33]:
df = df.sort_values(by=['experience_level', 'mouse_id'])
In [34]:
multi_session_df[multi_session_df.ophys_container_id==1018028218].experience_level.unique()
Out[34]:
array(['Familiar', 'Novel 1', 'Novel >1'], dtype=object)
In [35]:
df[df.ophys_container_id==1018028218].experience_level.unique()
Out[35]:
array(['Familiar', 'Novel 1', 'Novel >1'], dtype=object)
In [36]:
data = df[df.epoch!=6].copy() # get rid of weird last epoch
data = data.groupby(['experience_level', 'epoch', 'cell_type', 'mouse_id', ]).mean()[['mean_response']].unstack(level=[0,1])
data.columns = data.columns.droplevel()

by mouse_id

In [37]:
fig, ax = plt.subplots(figsize=(15,15))
ax = sns.heatmap(data, vmin=0, vmax=0.005, ax=ax)

by area and depth and mouse

In [38]:
df = df.sort_values(by=['experience_level', 'mouse_id', 'targeted_structure', 'depth', ])
In [39]:
np.sort(df.experience_level.unique())
Out[39]:
array(['Familiar', 'Novel 1', 'Novel >1'], dtype=object)
In [40]:
df[df.ophys_container_id==1018028218]
Out[40]:
index cell_specimen_id stimulus_change epoch mean_response sem_response mean_trace sem_trace trace_timestamps mean_responses ... area_depth area_binned_depth first_novel n_relative_to_first_novel last_familiar last_familiar_active second_novel second_novel_active experience_exposure experience_epoch
39205 39205 1086564116 True 0.0 0.027243 0.018453 [0.0, 0.0, 0.0, 0.0, 0.0, 0.06016995609714393,... [0.0, 0.0, 0.0, 0.0, 0.0, 0.05890307829563513,... [-4.941422091574936, -4.848187712488621, -4.75... [0.0, 0.0, 0.33680440442527027, 0.0, 0.0, 0.0,... ... VISl_289 VISl_275 False -4.0 False False False False Familiar 0 Familiar epoch 1
39206 39206 1086564116 True 1.0 0.000000 0.000000 [0.0, 0.1888737706547727, 0.0, 0.0, 0.0, 0.0, ... [0.0, 0.13242262307734362, 0.0, 0.0, 0.0, 0.0,... [-4.941422091574937, -4.848187712488619, -4.75... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... ... VISl_289 VISl_275 False -4.0 False False False False Familiar 0 Familiar epoch 2
39207 39207 1086564116 True 2.0 0.000000 0.000000 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0996912056430... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0957802122613... [-4.941422091574937, -4.848187712488619, -4.75... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... ... VISl_289 VISl_275 False -4.0 False False False False Familiar 0 Familiar epoch 3
39208 39208 1086564116 True 3.0 0.027224 0.026598 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [-4.941422091574936, -4.84818771248862, -4.754... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.59893644... ... VISl_289 VISl_275 False -4.0 False False False False Familiar 0 Familiar epoch 4
39209 39209 1086564116 True 4.0 0.051770 0.034516 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0438006586799... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0425665889542... [-4.941422091574937, -4.84818771248862, -4.754... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... ... VISl_289 VISl_275 False -4.0 False False False False Familiar 0 Familiar epoch 5
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
280143 280143 1086577999 True 1.0 0.000000 0.000000 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [-4.941524531767727, -4.848288219847579, -4.75... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... ... VISl_289 VISl_275 False 2.0 False False False False Novel 3 Novel >1 epoch 2
280144 280144 1086577999 True 2.0 0.000000 0.000000 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [-4.941524531767727, -4.848288219847579, -4.75... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... ... VISl_289 VISl_275 False 2.0 False False False False Novel 3 Novel >1 epoch 3
280145 280145 1086577999 True 3.0 0.000000 0.000000 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [-4.941524531767727, -4.84828821984758, -4.755... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... ... VISl_289 VISl_275 False 2.0 False False False False Novel 3 Novel >1 epoch 4
280146 280146 1086577999 True 4.0 0.000000 0.000000 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [-4.941524531767727, -4.848288219847579, -4.75... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... ... VISl_289 VISl_275 False 2.0 False False False False Novel 3 Novel >1 epoch 5
280147 280147 1086577999 True 5.0 0.000000 0.000000 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... [-4.941524531767727, -4.8482882198475785, -4.7... [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... ... VISl_289 VISl_275 False 2.0 False False False False Novel 3 Novel >1 epoch 6

1024 rows × 58 columns

In [41]:
data
Out[41]:
experience_level Familiar Novel 1 Novel >1
epoch 0.0 1.0 2.0 3.0 4.0 5.0 0.0 1.0 2.0 3.0 4.0 5.0 0.0 1.0 2.0 3.0 4.0 5.0
cell_type mouse_id
Excitatory 403491 0.001663 0.002378 0.001560 0.001729 0.002819 0.003757 NaN NaN NaN NaN NaN NaN 0.001898 0.001081 0.001050 0.001323 0.001401 0.001136
412366 0.005168 0.004827 0.005482 0.004693 0.004330 0.004204 0.005180 0.004643 0.004111 0.005397 0.004540 0.004661 0.004201 0.003509 0.002829 0.003036 0.003375 0.002662
425493 0.001815 0.001790 0.001717 0.001911 0.002247 0.002111 0.002017 0.001876 0.001676 0.001744 0.003332 0.003583 0.001836 0.002090 0.001578 0.002619 0.002435 0.002033
425496 0.001173 0.000852 0.000727 0.000654 0.000632 0.000577 0.003199 0.002295 0.002682 0.002446 0.002087 0.001753 0.001692 0.001051 0.000925 0.001090 0.001221 0.001126
431023 0.002899 0.001740 0.001404 0.001243 0.001033 0.000864 0.005256 0.003687 0.002907 0.002155 0.002259 0.002245 0.001899 0.001350 0.001203 0.001014 0.000920 0.000942
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
Vip Inhibitory 523922 0.019722 0.023676 0.017366 0.028259 0.018054 0.013572 0.030730 0.027051 0.021518 0.019497 0.016453 0.020861 0.009982 0.021960 0.023999 0.025587 0.012521 0.008411
523927 0.001049 0.000916 0.000484 0.000489 0.000667 0.000582 0.007070 0.004488 0.002980 0.001459 0.002542 0.005263 0.001145 0.000563 0.000220 0.000668 0.000331 0.001909
528097 0.007565 0.009342 0.011666 0.011392 0.017566 0.008799 NaN NaN NaN NaN NaN NaN 0.011040 0.005489 0.007738 0.007793 0.006658 0.009592
538219 0.001353 0.001907 0.002036 0.000741 0.001045 0.000648 0.008303 0.006014 0.003172 0.005519 0.003137 0.001326 0.001690 0.000610 0.000721 0.000718 0.000375 0.000462
544261 0.003675 0.002087 0.001170 0.000784 0.000256 0.000603 0.006603 0.004896 0.007660 0.011888 0.009363 0.008516 0.003853 0.003554 0.001755 0.002519 0.002757 0.003056

82 rows × 18 columns

In [42]:
data = df[df.epoch!=6].copy() # get rid of weird last epoch
data = data.groupby(['experience_level', 'epoch', 'cell_type', 'mouse_id', ]).mean()[['mean_response']].unstack(level=[0,1])
data.columns = data.columns.droplevel()
In [43]:
conditions = ['experience_level', 'epoch', 'targeted_structure', 'depth', 'mouse_id', ]
for cell_type in cell_types:
    data = df[df.epoch!=6].copy() # get rid of weird last epoch
    data = data[data.cell_type==cell_type]
    data = data.sort_values(by=conditions)
    data = data.groupby(conditions).mean()[['mean_response']].unstack(level=[0,1])
    data.columns = data.columns.droplevel()
    
    fig, ax = plt.subplots(figsize=(16,8))
    ax = sns.heatmap(data, vmin=0, vmax=np.percentile(df[df.cell_type==cell_type].mean_response.values, 95), ax=ax)
    ax.set_title(cell_type)
    ax.set_ylim(-0.5, data.shape[0]+0.5)
In [ ]:

by area and depth, average over mice

In [44]:
experiments_table[(experiments_table.mouse_id==482853)&(experiments_table.targeted_structure=='VISl')&
                 (experiments_table.depth==350)]
Out[44]:
equipment_name full_genotype mouse_id reporter_line driver_line sex age_in_days cre_line indicator session_number ... binned_depth area_depth area_binned_depth first_novel n_relative_to_first_novel last_familiar last_familiar_active second_novel second_novel_active experience_exposure
ophys_experiment_id

0 rows × 37 columns

In [45]:
for cell_type in cell_types:
    # avg over mice
    conditions = ['experience_level', 'epoch', 'targeted_structure', 'depth']
    data = df[df.epoch!=6].copy() # get rid of weird last epoch
    data = data[data.cell_type==cell_type]
    data = data.sort_values(by=conditions)
    data = data.groupby(conditions).mean()[['mean_response']].unstack(level=[0,1])    
    data.columns = data.columns.droplevel()
    
    fig, ax = plt.subplots(figsize=(16,4))
    ax = sns.heatmap(data, vmin=0, vmax=np.percentile(df[df.cell_type==cell_type].mean_response.values, 95), ax=ax)
    ax.set_title(cell_type)
    ax.set_ylim(-0.5, data.shape[0]+0.5)
    
    # all mice
    conditions = ['experience_level', 'epoch', 'targeted_structure', 'depth', 'mouse_id']
    data = df[df.epoch!=6].copy() # get rid of weird last epoch
    data = data[data.cell_type==cell_type]
    data = data.sort_values(by=conditions)
    data = data.groupby(conditions).mean()[['mean_response']].unstack(level=[0,1])
    data.columns = data.columns.droplevel()
    
    fig, ax = plt.subplots(figsize=(16,12))
    ax = sns.heatmap(data, vmin=0, vmax=np.percentile(df[df.cell_type==cell_type].mean_response.values, 95), ax=ax)
    ax.set_title(cell_type)
    ax.set_ylim(-0.5, data.shape[0]+0.5)

median across mice

In [46]:
conditions = ['experience_level', 'epoch', 'targeted_structure', 'depth']
for cell_type in cell_types:
    # avg over mice
    data = df[df.epoch!=6].copy() # get rid of weird last epoch
    data = data[data.cell_type==cell_type]
    data = data.sort_values(by=conditions)
    data = data.groupby(conditions).mean()[['mean_response']].unstack(level=[0,1])
    data.columns = data.columns.droplevel()
    
    fig, ax = plt.subplots(figsize=(16,4))
    ax = sns.heatmap(data, vmin=0, vmax=np.percentile(df[df.cell_type==cell_type].mean_response.values, 95), ax=ax)
    ax.set_title(cell_type)
    ax.set_ylim(-0.5, data.shape[0]+0.5)
In [ ]:

sort by mouse and project code first

In [47]:
conditions = ['experience_level', 'epoch', 'project_code', 'mouse_id', 'targeted_structure', 'depth']
for cell_type in cell_types:
    # all mice
    data = df[df.epoch!=6].copy() # get rid of weird last epoch
    data = data[data.cell_type==cell_type]
    data = data.sort_values(by=conditions)    
    data = data.groupby(conditions).mean()[['mean_response']].unstack(level=[0, 1])
    data.columns = data.columns.droplevel()
    
    fig, ax = plt.subplots(figsize=(16,12))
    ax = sns.heatmap(data, vmin=0, vmax=np.percentile(df[df.cell_type==cell_type].mean_response.values, 95), ax=ax)
    ax.set_title(cell_type)
    ax.set_ylim(-0.5, data.shape[0]+0.5)
In [ ]:

without epochs

In [48]:
df = multi_session_df.copy()

conditions = ['experience_level', 'targeted_structure', 'depth']
for cell_type in cell_types:
    # avg over mice
    data = df.copy() 
    data = data[data.cell_type==cell_type]
    data = data.sort_values(by=conditions)
    data = data.groupby(conditions).mean()[['mean_response']].unstack(level=[0])
    data.columns = data.columns.droplevel()
    
    fig, ax = plt.subplots(figsize=(5,4))
    ax = sns.heatmap(data, vmin=0, vmax=np.percentile(df[df.cell_type==cell_type].mean_response.values, 80), ax=ax)
    ax.set_title(cell_type)
    ax.set_ylim(-0.5, data.shape[0]+0.5)
In [49]:
%%time

# set various params
df_name = 'trials_response_df'
conditions = ['cell_specimen_id', 'stimulus_change', 'change_image_name']
use_events = True
filter_events = False

# load multi_session_df
multi_session_df = loading.get_multi_session_df(cache_dir, df_name, conditions, experiments_table, 
                                                use_events=use_events, filter_events=filter_events)
print(len(multi_session_df.ophys_experiment_id.unique()))

original_multi_session_df = multi_session_df.copy()

# limit to changes, pref stim
multi_session_df = multi_session_df[(multi_session_df.stimulus_change==True)&
                                   (multi_session_df.pref_stim==True)]

# limit to platform paper dataset
multi_session_df = multi_session_df[multi_session_df.ophys_experiment_id.isin(experiments_table.index.values)]
print(len(multi_session_df.ophys_experiment_id.unique()))
# merge with metadata
multi_session_df = multi_session_df.merge(experiments_table, on='ophys_experiment_id')
print(len(multi_session_df.ophys_experiment_id.unique()))

1941
1249
1249
Wall time: 4min 15s
In [50]:
for cell_type in cell_types: 
    df = multi_session_df.copy()
    df = df[df.cell_type==cell_type]

    for project_code in df.project_code.unique():
        fig, ax = plt.subplots(figsize=(15,5))
        ax = sns.pointplot(data=df[df.project_code==project_code], x='experience_level', y='mean_response', hue='mouse_id', 
                           order=np.sort(df.experience_level.unique()), ax=ax)
        ax.get_legend().remove()
        ax.set_title(cell_type+', '+project_code)
In [51]:
multi_session_df.keys()
Out[51]:
Index(['cell_specimen_id', 'stimulus_change', 'change_image_name',
       'mean_response', 'sem_response', 'mean_trace', 'sem_trace',
       'trace_timestamps', 'mean_responses', 'mean_baseline', 'sem_baseline',
       'pref_stim', 'peak_response', 'time_to_peak', 'fano_factor',
       'fraction_significant_p_value_gray_screen',
       'fraction_significant_p_value_omission',
       'fraction_significant_p_value_stimulus', 'reliability',
       'ophys_experiment_id', 'equipment_name', 'full_genotype', 'mouse_id',
       'reporter_line', 'driver_line', 'sex', 'age_in_days', 'cre_line',
       'indicator', 'session_number', 'prior_exposures_to_session_type',
       'prior_exposures_to_image_set', 'prior_exposures_to_omissions',
       'ophys_session_id', 'behavior_session_id', 'ophys_container_id',
       'project_code', 'imaging_depth', 'targeted_structure',
       'date_of_acquisition', 'session_type', 'experience_level', 'passive',
       'image_set', 'file_id', 'cell_type', 'depth', 'binned_depth',
       'area_depth', 'area_binned_depth', 'first_novel',
       'n_relative_to_first_novel', 'last_familiar', 'last_familiar_active',
       'second_novel', 'second_novel_active', 'experience_exposure'],
      dtype='object')
In [52]:
data
Out[52]:
experience_level Familiar Novel 1 Novel >1
targeted_structure depth
VISl 72 0.023609 0.023816 NaN
75 0.014027 0.068773 0.012071
83 0.008285 NaN 0.009857
142 0.013672 0.008609 0.002916
143 0.041487 0.037266 0.024078
146 0.005087 NaN 0.006770
149 0.015942 NaN 0.008300
150 0.018341 0.049827 0.023867
163 0.005140 NaN 0.006744
219 0.005142 NaN 0.004932
221 0.008939 0.014651 0.007981
222 0.007483 NaN 0.008447
224 0.004764 0.007944 0.003708
225 0.012613 0.027909 0.008149
275 0.007129 0.009569 NaN
280 0.008909 NaN 0.004287
297 0.004068 0.003573 0.002758
298 0.007527 0.004404 0.004218
300 0.004847 0.005616 0.005525
303 0.014688 0.016366 0.016015
VISp 71 0.006638 0.015817 0.004417
73 0.011328 NaN 0.005084
74 0.000523 0.014523 0.000896
75 0.012107 0.024662 0.012009
78 0.016568 0.018346 0.008833
79 0.007471 0.108448 0.018452
150 0.008769 0.038853 0.013367
151 0.011453 0.018866 0.007264
152 0.009886 NaN 0.005111
153 0.002764 0.046480 0.006576
155 0.013005 0.027858 NaN
175 0.001148 0.005348 0.001653
219 0.003876 NaN 0.003797
220 0.002678 0.019781 0.006905
223 0.006337 0.004216 0.011749
225 0.009578 0.018758 0.005044
227 0.035137 0.030365 0.015937
273 0.008321 0.013644 0.015858
301 0.006714 0.007881 0.003847
In [53]:
conditions = ['experience_exposure', 'project_code', 'mouse_id', 'targeted_structure', 'depth']
for cell_type in cell_types:
    # all mice
    data = multi_session_df.copy()
    data = data[data.cell_type==cell_type]
    data = data.sort_values(by=conditions)    
    data = data.groupby(conditions).mean()[['mean_response']].unstack(level=[0])
#     data.columns = data.columns.droplevel()
    
    fig, ax = plt.subplots(figsize=(16,12))
    ax = sns.heatmap(data, vmin=0, vmax=np.percentile(multi_session_df[multi_session_df.cell_type==cell_type].mean_response.values, 95), ax=ax)
    ax.set_title(cell_type)
    ax.set_ylim(-0.5, data.shape[0]+0.5)
In [ ]:
In [54]:
### i
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:

plot population average across experience levels for Multiscope experiments

TBD: interpolate to average scientifica & multiscope

In [ ]:
os.path.join(loading.get_platform_analysis_cache_dir(), 'multi_session_summary_dfs_250ms')
In [ ]:
%%time

# set various params
df_name = 'trials_response_df'
conditions = ['cell_specimen_id', 'stimulus_change']
use_events = True
filter_events = True

# load multi_session_df
multi_session_df = loading.get_multi_session_df(cache_dir, df_name, conditions, experiments_table, 
                                                use_events=use_events, filter_events=filter_events)
print(len(multi_session_df.ophys_experiment_id.unique()))

original_multi_session_df = multi_session_df.copy()

# limit to changes, pref stim
multi_session_df = multi_session_df[(multi_session_df.stimulus_change==True)]

# limit to platform paper dataset
multi_session_df = multi_session_df[multi_session_df.ophys_experiment_id.isin(experiments_table.index.values)]
print(len(multi_session_df.ophys_experiment_id.unique()))
# merge with metadata
multi_session_df = multi_session_df.merge(experiments_table, on='ophys_experiment_id')
print(len(multi_session_df.ophys_experiment_id.unique()))
In [ ]:
# set project code 
project_code = 'VisualBehaviorMultiscope'
In [ ]:
# get timestamps 
multiscope_expt = experiments_table[experiments_table.project_code==project_code].index.values[9] 
timestamps = ppf.get_timestamps_for_response_df_type(cache, multiscope_expt, df_name)
In [ ]:
# plot
axes_column = 'cell_type'
hue_column = 'experience_level'
palette = utilities.get_experience_level_colors()
xlim_seconds = [-1.8,2.25]

df = multi_session_df[multi_session_df.project_code==project_code]
df = df[df.experience_level!='Novel >1']

ppf.plot_population_averages_for_conditions(df, df_name, timestamps, 
                                            axes_column, hue_column, project_code=project_code,
                                            use_events=use_events, filter_events=filter_events, 
                                            xlim_seconds=xlim_seconds, palette=palette,
                                            horizontal=True, save_dir=None, folder=None)
In [ ]:
In [ ]:
In [ ]:
%%time

# set various params
df_name = 'trials_response_df'
conditions = ['cell_specimen_id', 'stimulus_change', 'change_image_name']
use_events = True
filter_events = True

# load multi_session_df
multi_session_df = loading.get_multi_session_df(cache_dir, df_name, conditions, experiments_table, 
                                                use_events=use_events, filter_events=filter_events)
print(len(multi_session_df.ophys_experiment_id.unique()))

original_multi_session_df = multi_session_df.copy()

# limit to changes, pref stim
multi_session_df = multi_session_df[(multi_session_df.stimulus_change==True)]

# limit to platform paper dataset
multi_session_df = multi_session_df[multi_session_df.ophys_experiment_id.isin(experiments_table.index.values)]
print(len(multi_session_df.ophys_experiment_id.unique()))
# merge with metadata
multi_session_df = multi_session_df.merge(experiments_table, on='ophys_experiment_id')
print(len(multi_session_df.ophys_experiment_id.unique()))
In [ ]:
# plot
project_code = 'VisualBehaviorMultiscope'
# get timestamps 
multiscope_expt = experiments_table[experiments_table.project_code==project_code].index.values[9] 
timestamps = ppf.get_timestamps_for_response_df_type(cache, multiscope_expt, df_name)

axes_column = 'cell_type'
hue_column = 'experience_level'
palette = utilities.get_experience_level_colors()
xlim_seconds = [-1.8,2.25]

df = multi_session_df[multi_session_df.project_code==project_code]
df = df[df.experience_level!='Novel >1']
df = df[df.pref_stim==True]

ppf.plot_population_averages_for_conditions(df, df_name, timestamps, 
                                            axes_column, hue_column, project_code=project_code,
                                            use_events=use_events, filter_events=filter_events, 
                                            xlim_seconds=xlim_seconds, palette=palette,
                                            horizontal=True, save_dir=None, folder=None)
In [ ]:
%%time

# set various params
df_name = 'omission_response_df'
conditions = ['cell_specimen_id']
use_events = True
filter_events = True

# load multi_session_df
multi_session_df = loading.get_multi_session_df(cache_dir, df_name, conditions, experiments_table, 
                                                use_events=use_events, filter_events=filter_events)
print(len(multi_session_df.ophys_experiment_id.unique()))

original_multi_session_df = multi_session_df.copy()

# limit to changes, pref stim
# multi_session_df = multi_session_df[(multi_session_df.stimulus_change==True)]

# limit to platform paper dataset
multi_session_df = multi_session_df[multi_session_df.ophys_experiment_id.isin(experiments_table.index.values)]
print(len(multi_session_df.ophys_experiment_id.unique()))
# merge with metadata
multi_session_df = multi_session_df.merge(experiments_table, on='ophys_experiment_id')
print(len(multi_session_df.ophys_experiment_id.unique()))
In [ ]:
# plot
project_code = 'VisualBehaviorMultiscope'

# get timestamps 
multiscope_expt = experiments_table[experiments_table.project_code==project_code].index.values[9] 
timestamps = ppf.get_timestamps_for_response_df_type(cache, multiscope_expt, df_name)

axes_column = 'cell_type'
hue_column = 'experience_level'
palette = utilities.get_experience_level_colors()
xlim_seconds = [-1.8,2.25]

df = multi_session_df[multi_session_df.project_code==project_code]
df = df[df.experience_level!='Novel >1']

ppf.plot_population_averages_for_conditions(df, df_name, timestamps, 
                                            axes_column, hue_column, project_code=project_code,
                                            use_events=use_events, filter_events=filter_events, 
                                            xlim_seconds=xlim_seconds, palette=palette,
                                            horizontal=True, save_dir=None, folder=None)
In [ ]:
%%time

# set various params
df_name = 'trials_response_df'
conditions = ['cell_specimen_id', 'stimulus_change']
use_events = False
filter_events = False

# load multi_session_df
multi_session_df = loading.get_multi_session_df(cache_dir, df_name, conditions, experiments_table, 
                                                use_events=use_events, filter_events=filter_events)
print(len(multi_session_df.ophys_experiment_id.unique()))

original_multi_session_df = multi_session_df.copy()

# limit to changes, pref stim
multi_session_df = multi_session_df[(multi_session_df.stimulus_change==True)]

# limit to platform paper dataset
multi_session_df = multi_session_df[multi_session_df.ophys_experiment_id.isin(experiments_table.index.values)]
print(len(multi_session_df.ophys_experiment_id.unique()))
# merge with metadata
multi_session_df = multi_session_df.merge(experiments_table, on='ophys_experiment_id')
print(len(multi_session_df.ophys_experiment_id.unique()))
In [ ]:
# plot
project_code = 'VisualBehaviorMultiscope'

# get timestamps 
multiscope_expt = experiments_table[experiments_table.project_code==project_code].index.values[9] 
timestamps = ppf.get_timestamps_for_response_df_type(cache, multiscope_expt, df_name)

axes_column = 'cell_type'
hue_column = 'experience_level'
palette = utilities.get_experience_level_colors()
xlim_seconds = [-1.8,2.25]

df = multi_session_df[multi_session_df.project_code==project_code]

ppf.plot_population_averages_for_conditions(df, df_name, timestamps, 
                                            axes_column, hue_column, project_code=project_code,
                                            use_events=use_events, filter_events=filter_events, 
                                            xlim_seconds=xlim_seconds, palette=palette,
                                            horizontal=True, save_dir=None, folder=None)
In [ ]:
In [ ]:
In [ ]:

plot population average with exposure number

In [ ]:
hue_column = 'prior_exposures_to_omissions'
axes_column= 'cell_type'
palette = sns.color_palette('magma',8)
xlim_seconds = [-1.8,2.25]

df = multi_session_df.copy()
df = df[df.experience_level=='Familiar']
df = df[df.prior_exposures_to_omissions<=6]

ppf.plot_population_averages_for_conditions(df, df_name, timestamps, 
                                            axes_column, hue_column, project_code=project_code,
                                            use_events=use_events, filter_events=filter_events, 
                                            xlim_seconds=xlim_seconds, palette=palette,
                                            horizontal=True, save_dir=None, folder=None)

plot population average across cell types

In [ ]:
axes_column = 'experience_level'
hue_column = 'cell_type'
palette = utils.get_cre_line_colors()
xlim_seconds = [-1.8,2.25]

ppf.plot_population_averages_for_conditions(df, df_name, timestamps, 
                                            axes_column, hue_column, project_code=project_code,
                                            use_events=use_events, filter_events=filter_events, 
                                            xlim_seconds=xlim_seconds, palette=palette,
                                            horizontal=True, save_dir=None, folder=None)

plot cell response heatmaps

In [ ]:
matched_cells = pd.read_csv(r"\\allen\programs\braintv\workgroups\nc-ophys\visual_behavior\platform_paper_cache\matched_cell_lists\cells_matched_across_experience_levels_full_dataset.csv")
matched_cell_ids = matched_cells.cell_specimen_id.unique()
In [ ]:
sdf = multi_session_df.copy()
# sdf = sdf[sdf.cell_specimen_id.isin(matched_cell_ids)]
sdf = sdf.sort_values(by='date_of_acquisition', ascending=True)
sdf = sdf.drop_duplicates(subset=['cell_specimen_id', 'experience_level'], keep='first') 

row_condition = 'cell_type'
col_condition = 'experience_level'
xlim_seconds = [(3-1.8)*11,(3+2.25)*11]

ppf.plot_response_heatmaps_for_conditions(sdf, df_name, timestamps,
                                        row_condition, col_condition, use_events, filter_events, project_code=project_code,
                                        microscope='Multiscope', vmax=0.04, xlim_seconds=xlim_seconds, 
                                        match_cells=True, save_dir=None, folder=None)
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]: